geney 1.4.30__py2.py3-none-any.whl → 1.4.32__py2.py3-none-any.whl

geney/utils/SeqMats.py

@@ -215,6 +215,8 @@ class SeqMat:
         #     else:
         #         raise ValueError(f"Unsupported mutation {pos}:{ref}:{alt}.")
         # Bucket mutations
+
+
         subs, ins, dels = [], [], []
         for p, r, a in mutations:
             # left-normalize
@@ -286,18 +288,30 @@ class SeqMat:
         self.notes['total_mutations'] = self.notes.get('total_mutations', 0) + applied_mutations
         return self
 
-    def complement(self) -> SeqMat:
-        comp = {b'A': b'T', b'T': b'A', b'C': b'G', b'G': b'C', b'-': b'-'}
-        nts = np.array([comp[x] for x in self.seq_array['nt']], dtype='S1')
-        new = self.clone()
-        new.seq_array['nt'] = nts
-        return new
+    # def complement(self) -> SeqMat:
+    #     comp = {b'A': b'T', b'T': b'A', b'C': b'G', b'G': b'C', b'-': b'-'}
+    #     nts = np.array([comp[x] for x in self.seq_array['nt']], dtype='S1')
+    #     new = self.clone()
+    #     new.seq_array['nt'] = nts
+    #     return new
+    #
+    # def reverse_complement(self) -> SeqMat:
+    #     new = self.complement().clone()
+    #     new.seq_array = new.seq_array[::-1].copy()
+    #     new.rev = not self.rev
+    #     return new
+    def complement(self) -> "SeqMat":
+        comp_dict = {b"A": b"T", b"T": b"A", b"C": b"G", b"G": b"C", b"-": b"-", b"N": b"N"}
+        comp_seq = np.array([comp_dict[nt] for nt in self.seq_array["nt"]], dtype="S1")
+        new_instance = self.clone()
+        new_instance.seq_array["nt"] = comp_seq
+        return new_instance
 
-    def reverse_complement(self) -> SeqMat:
-        new = self.complement().clone()
-        new.seq_array = new.seq_array[::-1].copy()
-        new.rev = not self.rev
-        return new
+    def reverse_complement(self) -> "SeqMat":
+        rev_comp_seq = self.complement().seq_array[::-1]
+        self.seq_array = rev_comp_seq.copy()
+        self.rev = not self.rev
+        return self
 
     def __getitem__(self, key: Union[int, slice]) -> np.ndarray:
         coords = self.seq_array['index']
@@ -424,7 +438,7 @@ class SeqMat:
 
         # Run the splicing prediction engine (function assumed to be defined externally)
         from .splicing_utils import run_splicing_engine
-        donor_probs, acceptor_probs = run_splicing_engine(seq, engine)
+        donor_probs, acceptor_probs = run_splicing_engine(seq=seq, splicing_engine=engine)
         # Trim off the fixed flanks before returning results.
         seq = seq[5000:-5000]
         indices = indices[5000:-5000]

geney/utils/spliceai_pytorch_utils.py

@@ -11,29 +11,44 @@ from spliceai_pytorch import SpliceAI
 model = SpliceAI.from_preconfigured('10k')
 
 
+device = torch.device('cpu')
 if sys.platform == 'darwin':
     device = torch.device("mps") if torch.backends.mps.is_available() else torch.device("cpu")
 
 if sys.platform == 'linux':
     device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
 
-
 print(f"SpliceAI loaded to {device}.")
 model.to(device)
 
-def one_hot_encode(seq):
 
-    map = np.asarray([[0, 0, 0, 0],
-                      [1, 0, 0, 0],
-                      [0, 1, 0, 0],
-                      [0, 0, 1, 0],
-                      [0, 0, 0, 1]])
+def one_hot_encode(seq: str) -> torch.Tensor:
+    """
+    One-hot encodes a nucleotide sequence into shape [L, 4] (A, C, G, T).
+    Unknowns (N or other) are mapped to all-zero vectors.
+    """
+    map = np.array([
+        [0, 0, 0, 0],  # index 0: unknown (N, etc.)
+        [1, 0, 0, 0],  # A
+        [0, 1, 0, 0],  # C
+        [0, 0, 1, 0],  # G
+        [0, 0, 0, 1],  # T
+    ], dtype=np.float32)
+
+    # Build mapping: ASCII values
+    ascii_seq = np.frombuffer(seq.upper().encode("ascii"), dtype=np.uint8)
 
-    seq = seq.upper().replace('A', '\x01').replace('C', '\x02')
-    seq = seq.replace('G', '\x03').replace('T', '\x04').replace('N', '\x00')
+    # A=65, C=67, G=71, T=84 → map A/C/G/T to 1/2/3/4; others to 0
+    code_map = np.zeros(128, dtype=np.uint8)
+    code_map[ord('A')] = 1
+    code_map[ord('C')] = 2
+    code_map[ord('G')] = 3
+    code_map[ord('T')] = 4
 
-    return map[np.fromstring(seq, np.int8) % 5]
+    indices = code_map[ascii_seq]  # shape [L]
+    onehot = map[indices]          # shape [L, 4]
 
+    return torch.tensor(onehot, dtype=torch.float32)
 
 def sai_predict_probs(seq: str, model) -> list:
     '''
@@ -50,10 +65,12 @@ def sai_predict_probs(seq: str, model) -> list:
     is the donor probability. These probabilities corresponds to the
     middel <L NTs> NTs of the input seq.
     '''
-    x = one_hot_encode(seq)[None, :]
+    x = one_hot_encode(seq)[None, :, :].transpose(1, 2)  # shape: [1, 4, L]
     y = model(x)
-    y = y[0, :, 1:].T
-    return y[0, :], y[1, :]
+    probs = torch.softmax(y, dim=1)  # shape: [1, 3, L]
+    acceptor_probs = probs[0, :, 1]       # [L]
+    donor_probs    = probs[0, :, 2]       # [L]
+    return acceptor_probs.tolist(), donor_probs.tolist()
 
 
 def run_spliceai_seq(seq, indices, threshold=0):

geney/utils/splicing_utils.py

@@ -3,7 +3,6 @@
 import pandas as pd
 from typing import List, Tuple, Optional
 
-# def run_splicing_engine(seq: Optional[str] = None, engine: str = 'spliceai') -> Tuple[List[float], List[float]]:
 def run_splicing_engine(seq: Optional[str] = None, engine: str = 'spliceai') -> Tuple[List[float], List[float]]:
     """
     Run the specified splicing engine to predict splice site probabilities on a sequence.
@@ -25,9 +24,10 @@ def run_splicing_engine(seq: Optional[str] = None, engine: str = 'spliceai') ->
 
     match engine:
         case 'spliceai':
-            # from geney.utils.spliceai_utils import sai_predict_probs, sai_models
-            # acceptor_probs, donor_probs = sai_predict_probs(seq, models=sai_models)
+            from geney.utils.spliceai_utils import sai_predict_probs, sai_models
+            acceptor_probs, donor_probs = sai_predict_probs(seq, models=sai_models)
 
+        case 'spliceai-pytorch':
             from geney.utils.spliceai_pytorch_utils import sai_predict_probs, model
             acceptor_probs, donor_probs = sai_predict_probs(seq, model=model)
 
@@ -41,7 +41,6 @@ def run_splicing_engine(seq: Optional[str] = None, engine: str = 'spliceai') ->
     return donor_probs, acceptor_probs
 
 
-
 def adjoin_splicing_outcomes(splicing_predictions, transcript=None):
     """
     Predicts splicing effect for multiple mutations and organizes the output as a multi-index DataFrame.

{geney-1.4.30.dist-info → geney-1.4.32.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geney
-Version: 1.4.30
+Version: 1.4.32
 Summary: A Python package for gene expression modeling.
 Home-page: https://github.com/nicolaslynn/geney
 Author: Nicolas Lynn

{geney-1.4.30.dist-info → geney-1.4.32.dist-info}/RECORD

@@ -37,17 +37,17 @@ geney/translation_initiation/tis_utils.py,sha256=AF3siFjuQH-Rs44EV-80zHdbxRMvN4w
 geney/translation_initiation/resources/kozak_pssm.json,sha256=pcd0Olziutq-6H3mFWDCD9cujQ_AlZO-iiOvBl82hqE,1165
 geney/translation_initiation/resources/tis_regressor_model.joblib,sha256=IXb4DUDhJ5rBDKcqMk9zE3ECTZZcdj7Jixz3KpoZ7OA,2592025
 geney/utils/Fasta_segment.py,sha256=weB5NJ65P0XiyAJCiCHx4T9sHC1pWLpuQeOy0B85gyg,11364
-geney/utils/SeqMats.py,sha256=PaUp6PMKYDYZ8RTodmKAmCa9ywHnkqSEqTjPoATr82k,17616
+geney/utils/SeqMats.py,sha256=-eoSKJFZR5OelhjXVkTIJbqYgp2xoWvgo8KWfKIBkRk,18197
 geney/utils/SeqMatsOld.py,sha256=syRU5DAuTh3xUfGW_qP9wlcBO5pHsG_y5PlrfXTIxUY,18502
 geney/utils/TranscriptLibrary.py,sha256=W1hv4Y8wRlmwTs3iFdn4_IqS-2suVDzZe4fwti2KbR4,2076
 geney/utils/__init__.py,sha256=-nJ-DMx1JzP-ZCe_QuQCeM0ZYIT_16jxoXDhUaO_4Oc,714
 geney/utils/mutation_utils.py,sha256=r-pHr56gEa5kh_DPX8MjFY3ZfYaOtyo4CUfJ5ZHlXPw,3243
 geney/utils/pangolin_utils.py,sha256=JQSPbWxdzqGFYfWQktkfLMaMSGR28eGQhNzO7MLMe5M,6162
-geney/utils/spliceai_pytorch_utils.py,sha256=PfMgaoG6ftWfqKZKc_JNqj5wqQRUR2B-4YF22-zNh1M,2079
+geney/utils/spliceai_pytorch_utils.py,sha256=wwBoT2utKZjjaWAUo11mSoympVK1vkNxAxcJvXAh8SM,2792
 geney/utils/spliceai_utils.py,sha256=VtrIbjyQxk_3lw86eWjftRYyal9OzxArJ0GV5u_ymTg,2721
-geney/utils/splicing_utils.py,sha256=Hn7conjcEb3Qc5MTN2DvnlZVxgn3U_IKqOAkGuIik6g,21052
+geney/utils/splicing_utils.py,sha256=Z0Z645fJhFnszpIueeMNaa-NMRLFFyJoRBR_yFbzrN0,20965
 geney/utils/utils.py,sha256=GXqlatNhix1akt3fburNzIwhiW9ZdCQSt2vmU80neyA,2370
-geney-1.4.30.dist-info/METADATA,sha256=_DIWzuIrV5xrRCN8VNgctibosZNMXMEVIRsX7d24WNc,990
-geney-1.4.30.dist-info/WHEEL,sha256=AHX6tWk3qWuce7vKLrj7lnulVHEdWoltgauo8bgCXgU,109
-geney-1.4.30.dist-info/top_level.txt,sha256=O-FuNUMb5fn9dhZ-dYCgF0aZtfi1EslMstnzhc5IIVo,6
-geney-1.4.30.dist-info/RECORD,,
+geney-1.4.32.dist-info/METADATA,sha256=SLfl8T6fgOlpb2Plf1iJxcaD-IXjclU4ZFcD3SWZTbM,990
+geney-1.4.32.dist-info/WHEEL,sha256=AHX6tWk3qWuce7vKLrj7lnulVHEdWoltgauo8bgCXgU,109
+geney-1.4.32.dist-info/top_level.txt,sha256=O-FuNUMb5fn9dhZ-dYCgF0aZtfi1EslMstnzhc5IIVo,6
+geney-1.4.32.dist-info/RECORD,,